new model

app.py

@@ -0,0 +1,61 @@
+import tensorflow as tf
+from tensorflow import keras
+from tensorflow.keras import Sequential
+from tensorflow.keras.layers import Dense, Flatten
+import matplotlib.pyplot as plt
+import gradio as gr
+import numpy as np
+#%matplotlib inline
+
+
+objt=tf.keras.datasets.mnist
+(X_train, y_train), (X_test,y_test)=objt.load_data()
+
+print(X_train.shape)
+
+print(y_train)
+
+for i in range(9):
+    plt.subplot(330+1+i)
+    plt.imshow(X_train[i])
+    plt.show()
+
+X_train=X_train/255.0
+X_test=X_test/255.0
+
+model=tf.keras.models.Sequential([Flatten(input_shape=(28,28)),
+                                        Dense(784,activation='relu'),
+                                        Dense(650,activation='relu'),
+                                        Dense(550,activation='relu'),
+                                        Dense(450,activation='relu'),
+                                        Dense(350,activation='relu'),
+                                        Dense(256,activation='relu'),
+                                        Dense(128,activation='relu'),
+                                        Dense(80,activation='relu'),
+                                        Dense(40,activation='relu'),
+                                        
+                                        Dense(10,activation=tf.nn.softmax)])
+
+model.compile(optimizer='adam',
+              loss='sparse_categorical_crossentropy',
+              metrics=['accuracy'])
+model.fit(X_train,y_train, epochs=10)
+model.save("keras_digit_temp.h5")
+test=X_test[0].reshape(-1,28,28)
+predicted=model.predict(test)
+print(predicted)
+
+def prdict_digit(img):
+    loaded_model = keras.models.load_model('keras_digit_temp.h5')
+    img_3d=img.reshape(-1,28,28)
+    img_resized=img_3d/255.0
+    pred_prob=loaded_model.predict(img_resized)
+    predicted_val=np.argmax(pred_prob)
+    return int(predicted_val)
+
+iface=gr.Interface(prdict_digit, inputs='sketchpad', outputs='label').launch()
+
+iface.launch(debug='true')
+    
+
+

requirements.txt

@@ -0,0 +1,17 @@
+tensorflow
+keras
+
+scikit_learn<1.3.0
+seaborn==0.12.2
+
+torchvision<0.15.2
+gradio
+
+#streamlit==1.26.0
+
+
+
+
+
+
+

test.py

@@ -0,0 +1,50 @@
+import gradio as gr 
+#import streamlit as st
+from sklearn.neural_network import MLPClassifier 
+import torchvision.datasets as datasets 
+import seaborn as sns 
+import pickle
+
+#dark mode seaborn 
+sns.set_style("darkgrid")
+
+mnist_trainset = datasets.MNIST(root='./data', train=True, download=True, transform=None)
+mnist_testset = datasets.MNIST(root='./data', train=False, download=True, transform=None)
+
+print(mnist_trainset.data.shape)
+print(mnist_testset.data.shape)
+print(mnist_trainset.targets.shape)
+print(mnist_testset.targets.shape)
+X_train = mnist_trainset.data
+y_train = mnist_trainset.targets
+X_test = mnist_testset.data
+y_test = mnist_testset.targets
+
+X_train = X_train.numpy() 
+X_test = X_test.numpy()
+y_train = y_train.numpy()
+y_test = y_test.numpy()
+
+X_train = X_train.reshape(60000, 784)/255.0
+X_test = X_test.reshape(10000, 784)/255.0
+
+#train the model 
+mlp = MLPClassifier(hidden_layer_sizes=(32,32))
+mlp.fit(X_train, y_train)
+
+#print the accuracies 
+print("Training Accuracy: ", mlp.score(X_train, y_train))
+print("Testing Accuracy: ", mlp.score(X_test, y_test))
+
+pickle.dump(mlp, open("digitmodel.sav", 'wb'))
+
+
+loaded_model = pickle.load(open("digitmodel.sav", 'rb'))  #loding saved model
+
+def predict(img):
+    img = img.reshape(1,784)/255.0
+    prediction = loaded_model.predict(img)[0]
+    return int(prediction)
+
+gr.Interface(fn= predict, inputs = "sketchpad", outputs ="label").launch()
+